Messages in this thread | | | From | Brian Gerst <> | Date | Sun, 14 Aug 2016 08:55:00 -0400 | Subject | Re: [PATCH v3 09/51] x86/dumpstack: fix x86_32 kernel_stack_pointer() previous stack access |
| |
On Sun, Aug 14, 2016 at 3:26 AM, Andy Lutomirski <luto@amacapital.net> wrote: > On Fri, Aug 12, 2016 at 7:28 AM, Josh Poimboeuf <jpoimboe@redhat.com> wrote: >> On x86_32, when an interrupt happens from kernel space, SS and SP aren't >> pushed and the existing stack is used. So pt_regs is effectively two >> words shorter, and the previous stack pointer is normally the memory >> after the shortened pt_regs, aka '®s->sp'. >> >> But in the rare case where the interrupt hits right after the stack >> pointer has been changed to point to an empty stack, like for example >> when call_on_stack() is used, the address immediately after the >> shortened pt_regs is no longer on the stack. In that case, instead of >> '®s->sp', the previous stack pointer should be retrieved from the >> beginning of the current stack page. >> >> kernel_stack_pointer() wants to do that, but it forgets to dereference >> the pointer. So instead of returning a pointer to the previous stack, >> it returns a pointer to the beginning of the current stack. >> >> Fixes: 0788aa6a23cb ("x86: Prepare removal of previous_esp from i386 thread_info structure") >> Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com> > > This seems like a valid fix, but I'm not sure I agree with the intent > of the code. ®s->sp really is the previous stack pointer in the > sense that the stack pointer was ®s->sp when the entry happened. > From an unwinder's perspective, how is: > > movl [whatever], $esp > <-- interrupt > > any different from: > > movl [whatever], $esp > pushl [something] > <-- interrupt > > Also, does x86_32 do this type of stack switching at all? AFAICS > 32-bit kernels don't use IRQ stacks in the first place. Do they? Am > I just missing the code that does it?
32-bit uses a software-based stack switch to run on the IRQ stack. See execute_on_irq_stack() in irq_32.c.
-- Brian Gerst
| |